Recent study shows that Xinhua during the pre-1949 period was both news service and intelligence agency -- e.g., combinng foreign broadcasts and publications for information needed by the CCP leadership.
This jives with some of my secondary source research, which noted that as of the late 1980s, Xinhua published Cankao Xiaoxi (Reference News), which translates from foreign publications important articles, without embellishment. This is a must-read publication for CCP officials, and, despite being restricted to party and government officials, had a circulation of 8 million in the late 1980s (Mass Media in China: The History and the Future. Ames: Iowa State University Press, 1989, p. 69)
I have also read in the Hong Kong press that the Hong Kong branch of the Xinhua News Agency was in fact an intelligence-gathering facility. I did not include references to this in my paper, as it was tangental to the focus of my paper.
Professor Kuhn also notes the following about my paper:
I am impressed by the power of statistical approach to generate hypotheses. In each policy turn, you cite reasons drawn from non-quantitatve sources for the particular configuration of variables. Your reasoning is plausible in view of the evidence you cite from non-numerical sources. I wonder, though, if there could be a way to use a finer medshed array of frequency counts to support one or more of these hypotheses?
This is something I have been thinking about for my ALM thesis -- can I or should I take a hybrid numerical/non-numerical approach to my content analysis, as I did in my study of references to Vietnam, Kampuchea, and overseas Chinese, or try a more strictly quantitative approach, as Prof. Kuhn suggests?
To recap, I did the following in my research paper for Prof. Kuhn's class:
1) Used simple frequency counts to draw out patterns in the data
2) Used a traditional historical approach to explain the patterns as well as analyze qualitatively the news articles themselves.
In retrospect, I don't think I could have done this study in a wholly quantitative manner. There were shifts in Xinhua coverage of the three variables that could not readily be drawn out by automated means, except by applying contrived filters in order to force a quantitative data set that supported my hypotheses.
One example that springs to mind is the shift in Xinhua coverage of the Vietnam refugee crisis from the summer of 1978 to the spring of 1978. In the summer of 1978, Xinhua's criticism of Vietnam on this issue was very direct, citing the words and actions of Chinese officials. However, in the spring of 1979, the criticism was very indirect -- Xinhua cited other countries' criticism of Vietnam. There is no clear linguistic indicator that I could have used to discern this shift using quantitative methods.
But there may be for another content analysis on a different set of variables within the Xinhua archive. More on this in the coming days ....